Localized NNR spectroscopy (MRS) of observable brain proton (1H) metabolites levels is currently used in many centers for non-invasive studies of tumors, stroke, HIV infection, multiple sclerosis, and Alzheimer' disease. To cover a significant portion of the brain; current 1H MRS techniques interleave 2-D slices, each measured separately by a spin-echo sequence. Unfortunately, interleaving is an inefficient use of the limited clinical examination time, and the spin-echo delays cause T1-weighting and J-coupling modulations of the signals. Consequently the observed metabolite levels depend on the particular localization scheme used and the sensitivity is relatively low considering that the variation in these pathologies are less than 30 percent. The applicants proposed to address the issues of low sensitivity and sequence-dependence by developing 3-D localization methods using combinations (hybrids) of two non-echo, multivoxel 1H MRS techniques: Chemical Shift Imaging (CSl) with Hadamard Spectroscopic Imaging (HS1). Three significant benefits could be realized. First, per equal measurement time, 3-D hybrids will result in a two to three fold gain in sensitivity over 2-D interleaved acquisition (depending on whether 4 or 8 slices are sought). Second, because the hybrids are non-echo, the J- coupling modulation artifacts and T2 losses will be minimal, yielding another 25 to 50 percent of sensitivity. Combined, the overall three to five fold increase will provide a dramatic boost to the reliability of metabolite level evaluation. Third, because the development will result in software to generate optimized pulse sequences, the methods could be implemented on any imager capable of producing shaped pulses. Since expensive, fast shielded-gradients are not intrinsically required, the proposed research will make multivoxel brain 1H MRS accessible to sites not yet equipped with them.